xfs
[Top] [All Lists]

Re: 2.4.9 /dev/nst1 "Bad file descriptor"

To: Jonathan Dill <dill@xxxxxxxxxxxx>
Subject: Re: 2.4.9 /dev/nst1 "Bad file descriptor"
From: Timothy Shimmin <tes@xxxxxxxxxxxxxxxxxxxxxxx>
Date: Wed, 31 Oct 2001 12:08:46 +1100
Cc: Linux XFS Mailing List <linux-xfs@xxxxxxxxxxx>
In-reply-to: <3BDEE801.C1727CCA@umbi.umd.edu>; from dill@umbi.umd.edu on Tue, Oct 30, 2001 at 12:48:49PM -0500
References: <3BDEE801.C1727CCA@umbi.umd.edu>
Sender: owner-linux-xfs@xxxxxxxxxxx
Hi Jonathan,

On Tue, Oct 30, 2001 at 12:48:49PM -0500, Jonathan Dill wrote:
> Since upgrading from 2.4.3 to the 2.4.9 prerelease kernels, long tape
> writes fail after 1-2 GB on a 20 GB native tape with "Bad file
> descriptor" on /dev/nst1.  
I'm not sure I understand what you are saying here (I often read
things the wrong way).
Are you saying that after you've done a number of tape writes
each of a certain buffer size (which would be ?), that
when you have in total written out around 1 to 2 Gb that you
get this error ?

[I initially was thinking you were referring to the problem of
 large tape block writes of 1-2Mb and you might need to change
 max_sg_segs for the st module as described in 
 cmd/xfsdump/doc/README.xfsdump]

Are there any messages in the system logs from the st driver ?
Were you using xfsdump ?

> I switched back to the 2.4.3 kernel and the
> tapedrive is working fine.  When I get caught up on my backups and have
> some time to play with it, I'll send some more info.  I'm just hoping
> that someone may have heard of a problem like this and know what causes
> it.
> 
> -- 
> "Jonathan F. Dill" (dill@xxxxxxxxxxxx)
> CARB IT Coordinator
> Experimental Support Site http://concept.umbi.umd.edu
> 

--Tim


<Prev in Thread] Current Thread [Next in Thread>